Data Wrangling with R

Aravind Hebbali

16 Exercise Solutions

17 Exercise Solutions

17.1 Introduction

Try every Your Turn before reading its solution — the struggle is the lesson. Worked solutions below are seeded for @dplyr-basics; solutions for the remaining chapters are open stubs. To contribute one, open an issue or pull request with the chapter, the question, and a runnable chunk in this book’s style (|>-first, same data as the chapter).

We will use the following R packages:

library(dplyr)
library(readr)

17.2 dplyr Basics

ecom <-
  read_csv('https://raw.githubusercontent.com/rsquaredacademy/datasets/master/web.csv',
    col_types = cols_only(device = col_factor(levels = c("laptop", "tablet", "mobile")),
      referrer = col_factor(levels = c("bing", "direct", "social", "yahoo", "google")),
      purchase = col_logical(), n_pages = col_double(), n_visit = col_double(),
      duration = col_double(), order_value = col_double(), order_items = col_double()
    )
  )

Q1. What is the average number of pages visited by purchasers and non-purchasers?

ecom |>
  summarise(mean_pages = mean(n_pages), .by = purchase)

# A tibble: 2 × 2
  purchase mean_pages
  <lgl>         <dbl>
1 FALSE          4.79
2 TRUE          15.8

Purchasers browse far more pages (~16 vs ~5) — the signal behind the AOV case study.

Q2. What is the average time on site for purchasers vs non-purchasers?

ecom |>
  summarise(mean_time = mean(duration), .by = purchase)

# A tibble: 2 × 2
  purchase mean_time
  <lgl>        <dbl>
1 FALSE         355.
2 TRUE          359.

Time on site barely differs (~355 vs ~359 seconds): page depth, not duration, separates buyers here.

Q3. What is the average number of pages visited by purchasers and non-purchasers using mobile?

ecom |>
  filter(device == "mobile") |>
  summarise(mean_pages = mean(n_pages), .by = purchase)

# A tibble: 2 × 2
  purchase mean_pages
  <lgl>         <dbl>
1 FALSE          4.99
2 TRUE          16.2

17.3 Open stubs

Solutions for these chapters are not yet written — contributions welcome (see above):

@joining-tables-in-r-dplyr: no Your Turn block; propose join puzzles on customer/order
@dplyr-helper-functions: sampling/slicing drills on ecom
@r-pipe-magrittr: rewrite-a-chain exercises (%>% → |>, _, \(x))
@tibbles-in-r: tibble vs data.frame edge cases
@tidying-data-with-tidyr: the three Your Turn prompts in that chapter
@strings-in-r: URL/email parsing variants on mockstring
@date-and-time-in-r: the nine Your Turn blocks (intervals, TZ/DST, formats)
@categorical-data-in-r: the seven Your Turn blocks (lumping, reordering, recoding)

--- title: "Exercise Solutions" description: "Data Wrangling in R: worked solutions to chapter exercises, plus open stubs for contributors." --- # Exercise Solutions {#exercise-solutions} ## Introduction Try every **Your Turn** before reading its solution — the struggle is the lesson. Worked solutions below are seeded for @dplyr-basics; solutions for the remaining chapters are open stubs. To contribute one, open an issue or pull request with the chapter, the question, and a runnable chunk in this book's style (`|>`-first, same data as the chapter). We will use the following R packages: ```{r sol_libs, message=FALSE} library(dplyr) library(readr) ``` ## dplyr Basics ```{r sol_data, message=FALSE} ecom <- read_csv('https://raw.githubusercontent.com/rsquaredacademy/datasets/master/web.csv', col_types = cols_only(device = col_factor(levels = c("laptop", "tablet", "mobile")), referrer = col_factor(levels = c("bing", "direct", "social", "yahoo", "google")), purchase = col_logical(), n_pages = col_double(), n_visit = col_double(), duration = col_double(), order_value = col_double(), order_items = col_double() ) ) ``` **Q1. What is the average number of pages visited by purchasers and non-purchasers?** ```{r sol_dplyr1} ecom |> summarise(mean_pages = mean(n_pages), .by = purchase) ``` Purchasers browse far more pages (~16 vs ~5) — the signal behind the AOV case study. **Q2. What is the average time on site for purchasers vs non-purchasers?** ```{r sol_dplyr2} ecom |> summarise(mean_time = mean(duration), .by = purchase) ``` Time on site barely differs (~355 vs ~359 seconds): page depth, not duration, separates buyers here. **Q3. What is the average number of pages visited by purchasers and non-purchasers using mobile?** ```{r sol_dplyr3} ecom |> filter(device == "mobile") |> summarise(mean_pages = mean(n_pages), .by = purchase) ``` ## Open stubs Solutions for these chapters are not yet written — contributions welcome (see above): - @joining-tables-in-r-dplyr: no **Your Turn** block; propose join puzzles on `customer`/`order` - @dplyr-helper-functions: sampling/slicing drills on `ecom` - @r-pipe-magrittr: rewrite-a-chain exercises (`%>%` → `|>`, `_`, `\(x)`) - @tibbles-in-r: tibble vs data.frame edge cases - @tidying-data-with-tidyr: the three **Your Turn** prompts in that chapter - @strings-in-r: URL/email parsing variants on `mockstring` - @date-and-time-in-r: the nine **Your Turn** blocks (intervals, TZ/DST, formats) - @categorical-data-in-r: the seven **Your Turn** blocks (lumping, reordering, recoding)